- From: T.V Raman <raman@google.com>
- Date: Tue, 1 May 2007 11:14:02 -0700
- To: howcome@opera.com
- Cc: raman@google.com, dbaron@dbaron.org, public-html@w3.org
Hakon, Apologies, but you're being a bit too glib here and in the process papering over quite a few pigs:-) To clarify: Håkon Wium Lie writes: > Also sprach T.V Raman: > > > Why aren't we defining Javascript the same way as what you > > describe --i.e. make every failing program "somehow work". > > Why aren't we even defining CSS that way i.e. "somehow make every > > CSS rule parse and mean something." > > Why is HTML special? > > CSS was designed with error recovery built into the syntax. If > an Not to the same extent that is being demanded of HTML. Error recovery at the level of "unknown CSS property" should translate to "unknown attribute" or "unknown element" in HTML. It should not translate to the kind of tag soup we're talking of fixing up and blessing as the new-fangled HTML for the Web. > unknown property or unit is used, the CSS specification describes how > to handle it. Draconian error recovery for CSS to match what is proposed for HTML would be a *lot* more -- and I'm sure you dont want to go there:-) > > HTML specification didn't have this in the past. (XHTML tried > to Again, this is a second pig that is getting papered over:-) It has been stated many times here that HTML was under-specified. I believe that: A) HTML (for presentational purposes) was *never* specified, since the hope was that presentation would be left to user agents. But Web sites went ahead and used it as a presentational language anyway, and tagsoup browsers of the 90's gave authors little help with respect to doing otherwise, the rest is history. B) CSS provided a glimmer of hope with respect to separation of presentation from content, but the underlying tagsoup mess that it had to style, combined with lack of interoperable implementations has meant that that dream has never been fully (or even partially) realized. C) Finally, (and for the record I asserted this at the time in 2000) the transition to XHTML was botched by trying to compartmentalize XHTML into a completely separate mime-type -- again this is now history and water under the bridge. > address it by adding a draconian rule: stop processing once I believe the draconian rule from XML that you state would have not caused a problem if the transition from HTML to XHTML had followed the roadmap that was proposed by many of us in 1998-99 -- namely, 0) Define error recovery for malformed HTML4; 1) Implement that error recovery initially in a script tag that is placed in the document head while delivering old content to new browsers; you find > an error. Authors didn't comply and 2) Once those script-implemented fixups have been proven, move their implementation into the browser for speed, 4) While creating new content that is clean and well-formed AKA XHTML. See http://www.w3.org/MarkUp/future/papers/adobe-19980427.html >implementations soon started > defecting). Some of us think error handling should be added to HTML as > well, just like it So let's make sure we're talking the *same kind* of error handling. exists in CSS. From this perspective, HTML isn't > special: it, too, needs graceful error handing. Even if the markup > doesn't deserve it :-) Again, graceful error handling is one thing; but let's make sure that the thing doesn't turn into a complete disgrace;-) > > -h&kon > Håkon Wium Lie CTO °þe®ª > howcome@opera.com http://people.opera.com/howcome -- Best Regards, --raman Title: Research Scientist Email: raman@google.com WWW: http://emacspeak.sf.net/raman/ Google: tv+raman GTalk: raman@google.com, tv.raman.tv@gmail.com PGP: http://emacspeak.sf.net/raman/raman-almaden.asc
Received on Tuesday, 1 May 2007 18:14:39 UTC